A

acpsamw©

K

1

reiwnstsbaoehoe

0d

Colloids and Surfaces A: Physicochem. Eng. Aspects 298 (2007) 192–200

Study of suspension settling: A approach to determinesuspension classification and particle interactions

R. Vie a, N. Azema a,∗, J.C. Quantin a, E. Touraud b, M. Fouletier a

a Centre des Materiaux de Grande Diffusion, Ecole des Mines d’Ales, 6 Avenue de Clavieres, 30319 Ales Cedex, Franceb Laboratoire du Genie de l’Environnement Industriel, Ecole des Mines d’Ales, 6 Avenue de Clavieres, 30319 Ales Cedex, France

Received 18 April 2006; received in revised form 28 September 2006; accepted 26 October 2006Available online 12 November 2006

bstract

The properties of mineral suspensions do not evolve linearly with particle mass fraction. For some critical concentrations these properties arebruptly modified due to a change of relative interparticle interaction force prevalence. Using kaolin and glass beads as model materials, physico-hemical stability and rheological measurement have been carried out in order to propose a global and discriminant parameter enabling the nature ofarticle interactions to be identified. Different types of suspensions were studied namely: diluted suspensions, concentrated suspensions and soliduspensions. Physico-chemical stability, studied by optical dispersion analysis (Turbiscan MA 2000) enables the clarification rate to be measured

nd to propose a new global parameter, the “phase separation index”. In addition, rheological analysis was realized by the means of viscous floweasurement. Behaviour of solid suspensions is satisfactorily described by Oswald de Waele power law in the whole range of concentration studiedhich tends to demonstrate that the interactions remains the same in this range. 2006 Elsevier B.V. All rights reserved.

harac

vmsczbptc

iafl

eywords: Particle interactions; Classification; Suspension; Kaolin; Settling; C

. Introduction

Pastes and suspensions are widely used in industrial area,ecycling and waste processing. Stability and mechanical prop-rties which mainly depend on particle size, particle shape,nterparticle forces and solid concentration, are of practical asell as fundamental interest. The properties of these media doot evolve linearly with solid concentration. A few scientistsuch as Coussot and Ancey [1] and Tradros [2] distinguishedhree types of suspensions: diluted suspensions, concentrateduspensions and solid suspensions. This typology is mainlyased on the correlation between the prevalent interparticle inter-ctions and the corresponding rheological properties. In the casef supra-colloidal particles (from 1 to 100 �m) which are largenough to neglect Brownian diffusion, each type of suspension

as a characteristic settling behaviour governed by specific typesf interactions. In a diluted suspension, particles are far fromach other and have no colloidal interactions. They settle indi-

∗ Corresponding author. Tel.: +33 4 66 78 53 56; fax: +33 4 66 78 53 65.E-mail address: nathalie.azema@ema.fr (N. Azema).

2

2

pfi

927-7757/$ – see front matter © 2006 Elsevier B.V. All rights reserved.oi:10.1016/j.colsurfa.2006.10.074

terisation

idually [3]. In a concentrated suspension, particles settle as aore or less consolidated mass with a sharp boundary between

ettling suspension and supernatant. The suspension behaviourorresponds to a regime known as hindered settling (also calledone settling or mass settling [3,4]) and water moves in spacesetween particles. On the other hand, in solid suspensions, thearticles are in contact and settle under compression. Underhe influence of gravity the contact structure moves to a moreompact one [4,5].

This work proposes a new approach to determine particlenteractions through the study of the settling kinetics usingn optical analysis. Suspensions containing a high solid massraction that cannot be optically analysed were studied by rheo-ogical measurements.

. Materials and methods

.1. Kaolin suspensions

Experiments were carried out on kaolin from Prolabo Com-any (CAS number 1332-58-7) dispersed in demineralised andltered water in order to control chemical entities in the medium.

R. Vie et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 298 (2007) 192–200 193

fracto

StTsm

2

pPaS

c(i[

pw

2

Lrr2

2

sf

2

fp

aweoprsmpt

Fig. 1. X-ray dif

amples were prepared from kaolin dry powder to which dis-illed water was added until a total weight of 30 g was reached.he resulting suspensions undergo a 15 min stiring at a con-tant speed in order to obtain a homogenous and dispersededium.

.1.1. X-ray analysis of kaolin powderIn order to characterize the mineralogical composition of the

owder, X-ray diffraction analysis was carried out using a PhilipsW 1700 instrument. A copper cathode was used. The evalu-tion program used was: DIFFRACT-AT programmed by M.ocabim.

The X-ray diffractogram represented in Fig. 1 indi-ates the presence of another kind of clay than kaoliniteAl2Si2O5(OH)4), the illite. The approximate formula of whichs (K0.88Al2(Si3.12Al0.88)O10(OH)2 according to Rosenberg6].

The ratio illite/kaolinite was determined using the methodroposed by Larray based on X-rays diffraction [7]. This ratioas estimated to be of the order of 2 ± 1%.

.1.2. Size distribution analysis

Particle size distribution of the kaolin was studied using a

S 230 laser granulometer (Beckman-Coulter). Granulomet-ic distribution shows three modes at 0.3, 5.5 and 13 �m asepresented in Fig. 2 and a granular spreading (d95–d5) of2.6 �m.

2

d

Fig. 2. Particle size dist

gram of kaolin.

.2. Glass microbeads

In order to get an example of non-cohesive particle suspen-ions, suspensions based on glass microbeads of various sizesrom Potters Europe Society were studied.

.3. Multiple light scattering analysis

The variation of suspension stability as a function of massraction was studied at room temperature (21 ◦C) using the dis-ersion optical analyser Turbiscan MA 2000 (formulaction) [8].

The sample was introduced in a glass cylindrical cell andnalysed using a light beam emitted in near infrared (850 nmavelength) which scans the sedimentation column. The time

lapsed between two consecutive scans is chosen as a functionf the sedimentation kinetics of the suspension. For exam-le, in the case of concentrated suspensions, the scans wereecorded each minute. In the case of diluted suspensions, thecans were recorded each 15 min. Each scans lasts 20 s. Trans-itted and backscattered photons are analysed using detectors

laced respectively at 0◦ and 135◦ from the incident beam direc-ion.

.4. Strain-controlled rheometry

Rheological properties at room temperature (21 ◦C) wereetermined using a rotational ARES rheometer from TA Instru-

ribution of kaolin.

1 hysic

mCtmo

rh

3

3n

3

tfiTctttprbnFaccfiv

tlmo

iTdhics

P

aobpi

B

λ

w(

94 R. Vie et al. / Colloids and Surfaces A: P

ents Company. Parallel geometry was employed. According tooussot and Ancey [1] and Papo et al. recommendations [9], all

he measurements were carried out using 50 mm diameter alu-inium serrated plates in order to prevent suspension slipping

n the wall.All the suspensions were initially manually sheared at a high

ate while 1 min in order to suppress the previous rheologicalistory of the sample tested.

. Results and discussion

.1. Characterisation of kaolin suspensions: definition of aew global parameter

.1.1. Classification of kaolin suspensionsSpectra represented in Figs. 3 and 4 correspond respectively

o diluted and concentrated suspensions (the sedimentation pro-le is schematically depicted in the lower part of each figure).he spectrum of a solid suspension is similar to that of a con-entrated one. X-axis represents height of the tube and Y-axishe transmitted or backscattered light percentage. The values inhe column on the right part give the correspondence betweenhe recorded trace and the time elapsed. From these spectra it isossible to obtain classical determinations such as clarificationate. To do so, evolution of the clarification signal thickness haseen followed as a function of time. Clarification signal thick-ess was measured at 2% of transmitted light (Figs. 3 and 4).ig. 5 represents evolution of the clarification signal thickness asfunction of time for two types of suspensions (diluted and con-

entrated). Clarification rate was obtained from the slope of theurves. Points between 0 and 3 min were ignored because clari-cation front is not well formed at these times. When the settlingelocity was not constant during the measurement (especially in

t

rF

Fig. 3. Example of Turbiscan MA 2000

ochem. Eng. Aspects 298 (2007) 192–200

he case of high sedimentation rate) it was evaluated from theinear part of the curve observed in the beginning of the measure-

ent when particle movements are not disturbed by the increasef the concentration.

Nevertheless a new parameter named “phase separationndex” (PSI) was proposed in a global characterisation approach.his parameter is defined from the sedimentation study. It isefined (Eq. (1)) as the ratio between the sedimentation columneight (Hc) and the sediment height at given sedimentation time,.e. 15 min (Hs) multiplicated by the average value of the per-entage of backscattered light (Bs%) in the zone defined as theediment (i.e. in Hs):

SI =(

Hs

Hc

)× (Bs %) (1)

PSI is a dimensionless parameter. Hc and Hs were determineds indicated in Figs. 3 and 4. PSI takes into account the quantityf matter which is present in sediment because Bs%, the averageackscattering percentage of the sediment, is directly linked toarticle volume fraction (φ) and the particle size (d) [8] as shownn Eqs. (2) and (3):

s ∼ 1

(λ∗)1/2 (2)

∗(d, Φ) = 2d

[3φ(1 − g) Qs](3)

hereλ* is the photon transport length in the considered mediumm), Qs the scattering efficiency factor (dimensionless), and g is

he asymmetry factor (dimensionless) [8].

The evolutions of the PSI parameter and of the clarificationate as functions of mass ratio in suspensions are represented inig. 6.

spectrum of diluted suspensions.

R. Vie et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 298 (2007) 192–200 195

ectru

upaessta

3mamfWiBn

3ttccitpaobc

3m

Fig. 4. Example of Turbiscan MA 2000 sp

Clarification rate is generally slow (less than 0.5 mm/min) butndergoes an important increase when the solid fraction is com-rised between 4 and 9%. The clarification rate may reach valuess high as 5.2 mm/min. Phase separation index exhibits linearvolution with two discontinuities characterised by a change oflope. Discontinuities do not occur abruptly, so PSI makes pos-ible three zones A, B, and C to be distinguished correspondingo the different types of suspensions, i.e. diluted, concentratednd solid suspensions and two intermediate zones.

.1.1.1. Zone A: diluted suspensions (from 0.26 to 1.04% inass fraction). In this part, clarification rate is small and is

lmost independent of the kaolin concentration. So, in thisass fraction domain, particles do not interact through colloidal

orces. This behaviour is typical of the diluted suspensions [1,2].

hile settling processes, a transpararency gradient is observed

n the column which corresponds to a concentration gradient.ut even in the bottom of the tube increase in concentration isot high enough to enable colloidal interactions to occur.

test

Fig. 5. Determination of clarification rate for a diluted and a conc

m of concentrated and solid suspensions.

.1.1.2. Intermediate zone 1 (from 1.8 to 3.5% in mass frac-ion). In this domain, clarification rate remains small (lesshan 0.5 mm/min) and almost independent of the particle con-entration but phase separation index increases rapidly. Theoncentration gradient above-mentioned implies that changesn interparticular force prevalence may occur at these concen-rations. Particles are sufficiently close to each other to makeossible colloidal interactions to take place. Some agglomer-tes are formed, then, kinetics of sedimentation in the lower partf the tube is enhanced. The sediment height will be higher tooecause loose agglomerates occupy a greater volume than itsonstitutive particles due to their fractal structures [10].

.1.1.3. Zone B: concentrated suspensions (from 4 to 9% inass fraction). Zone B is characterized by an abrupt increase in

he clarification rate. Kaolin particles immediately form agglom-rates due to physico-chemical interactions. Forming largetructures, the agglomerated particles got a higher sedimenta-ion rate [3]. Moreover, the slope of the PSI variation versus

enterated suspension (0.18 and 3.8 mm/min, respectively).

196 R. Vie et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 298 (2007) 192–200

F s fraci (5) so

caTsmtsc[

3IPtna[ah

3tcowpitp

3

irfTasamisaiF

dppimestB

ig. 6. Clarification velocity and phase separation index as a function of masntermediate state 1, (3) concentrated suspensions, (4) intermediate state 2, and

oncentration curve is steeper because the sedimentation mech-nism changes radically and “hindered settling” regime occurs.his regime is characterized by a sharp boundary between theuspension and the supernatant [11]. The solids subside as aore or less consolidated structure due to the fact that concen-

ration is large enough to develop agglomerate contact as theyettle [12]. Water movement is hindered by solid phase thenlarification velocity decreases as a function of mass fraction13].

.1.1.4. Intermediate zone 2 (from 9.5 to 16% in mass fraction).n this zone, clarification velocity is strongly slowed down andSI evolution is disturbed. The contacts which occur between

he agglomerates while sedimentation takes place become moreumerous. In fact, by increasing the concentration of solids,gglomerates cannot grow indefinitely without interpenetration14,15]. The agglomerates are then smaller and able to formmore compact structure corresponding to a smaller sedimenteight which explains that PSI evolution is slowed down.

.1.1.5. Zone C: solid suspensions (from 16 to 40% in mass frac-ion). In this domain, all the particles (or agglomerates) form aontact network in the suspension and settling phenomenon canccur only by compression. Sedimentation rates tend to zero, theeight of the particles in the sample is too small to create a com-

ression force high enough to generate a significant rammingn the suspension which is already highly concentrated. Evolu-ion in the PSI is then due to the increase in the backscatteringercentage value generated by the higher initial concentration.

sibt

tion of model and industrial kaolin suspensions: (1) diluted suspensions, (2)lid suspensions.

.1.2. Kaolin suspensions from ceramics industryIn Fig. 6 the evolution of the phase separation index of

ndustrial samples which belong to ceramics industry is alsoepresented. Samples A(1,2,3) and B(1,2,3) are suspensions maderom two different kaolins formulated with different additives.he different numbers correspond to differences in formulationnd additive concentrations. The lower viscosity of industrialamples is due to their formulation; the addition of dispersinggents made possible measurements to be realised at higherass fraction (55–70%). The PSI values seem to be mainly

nfluenced by the type of kaolin. This parameter makes pos-ible the suspensions made from the two different kaolins And B to be differentiated. Clarification velocities correspond-ng to these samples tend to zero and were not represented inig. 6.

Unlike sedimentation velocity, PSI possesses significant andiscriminant values in the whole range of mass fractions sus-ension studied. In the case of model suspensions and the PSIlateau value makes possible to distinguish highly concentratedndustrial suspensions made from two different kaolins. Further-

ore, this parameter exhibits intermediate behaviour betweenach type of suspensions. This observation suggests that the tran-ition from a type of settling to another one is not abrupt, butakes place over a narrow range of concentration as asserted by. Fitch in the case of clarification/hindered settling regime tran-

ition [11]. This difference is due to the fact that clarification rates measured in the top of the tube while PSI is determined in theottom where concentration can increase while sedimentationakes place.

R. Vie et al. / Colloids and Surfaces A: Physico

t

3m

tafl

rtssebis

o

tf

η

wγ

s

la

tNvra

tfi

3o

sstaeET6id

Fig. 7. Flow measurements of kaolin suspensions.

So PSI appears to be more effective than clarification rate inhe description of the destabilisation process of suspensions.

.2. Study of solid suspensions using viscous floweasurements

In order to study solid suspensions with a mass fraction higherhan 40% a study of rheological properties of kaolin suspensionss a function of mass fraction was carried out through viscousow measurements (Fig. 7).

Mass fractions between 30 and 55% were studied. In thisange of concentrations suspension viscosities are too lowo be correctly analysed by the detector. Beyond this range,uspensions undergo fracturing phenomenon which makes mea-urement impossible. As it can be seen in Fig. 7 this disturbingffect occurs at 50% of mass fraction at the last point and 55%ut only beyond 100 s−1. The major part of these two rheograms

s still correct; then, the analysis of these samples remains pos-ible.

All the studied suspensions are shear-thinning, the evolutionf their viscosity as a function of shear rate obeys a power law

csda

Fig. 8. Coefficient of consistency as a function

chem. Eng. Aspects 298 (2007) 192–200 197

hat can be fitted with the Oswald de Weale model using theollowing equation (Eq. (4)):

= K(γ)n−1 (4)

here η is the viscosity (Pa s), K the consistency index (Pa sn),˙ the shear rate (s−1), and n is the flow index (0 < n < 1 forhear-thinning fluids; dimensionless).

This relation between shear rate and viscosity is typicallyinked to progressive destruction of fractal structured agglomer-tes [10,16] induced by shearing [17,18].

Consistency index (K) evolves exponentially with mass frac-ion (Fig. 8). Flow index (n) which indicates the departure fromewtonian behaviour does not exhibit a significant evolution, itsalue is 0.28 ± 0.05 for all the samples, so the straight line likeheograms (on a bi-logarithmic scale) in Fig. 7 can be considereds having the same slope.

No discontinuities have been noticed in the evolution of thesewo parameters in the Oswald de Waele power law with massraction changes. These results lead to the conclusion that theres no change in the suspension type in this range of concentration.

.3. Observation of the agglomeration state by microscopicbservations

Existence of different discontinuities in the evolution of PSIuggests that particle organisation is different in each type ofuspensions. In order to observe these structural differences,hree suspensions have been deposited on an aluminium surface,ir dried and observed using a high resolution environmentallectronic microscope Quanta 200 FEG from FEI Company.ach of them belongs to one of the three types of suspensions.heir mass fractions are respectively 1% for the diluted one,.3% for the concentrated one and 22.5% for the solid one. Ast can be seen in Fig. 9 the state of agglomeration after drying isifferent in each type of suspension.

Edges of the particles from the diluted suspensions appear

learly and it is possible to observe the stacking of plate-likehape of kaolin particles as it can be done with samples ofry powder before adding water. This is an indication that thegglomeration phenomenon is almost non-existent in this type

of mass fraction of kaolin suspensions.

198 R. Vie et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 298 (2007) 192–200

differ

osasaospWc[

3

ass

3

tsw

sw

isTz

siepip

aatbA

Fig. 9. MEB micrographs of the

f suspensions. In the case of concentrated suspensions, it is pos-ible to observe agglomerates having various sizes (between 40nd 90 �m). It can be inferred that all particles while in suspen-ion have been interacting with their neighbours. Small particlesre agglomerated with bigger ones, which explains why edgesf the particles are more difficult to identify. The structure of theolid suspension is comparable with that of concentrated sus-ensions, but agglomerates are smaller (between 15 and 60 �m).hile agglomerates are formed in the suspension, they enter in

ontact, due to high concentration and hinder their own growing13,14].

.4. Influence of particle interactions on PSI variations

In order to identify the influence of the cohesive particle inter-ction strength on the evolution of PSI, non-cohesive glass beadsuspensions and kaolin suspensions at different pH have beentudied.

.4.1. Influence of the pH on kaolin suspensions

As it has been previously asserted the settling process of

he suspensions depends on the interparticle interactions. Thetrength of these different interactions can be deeply modifiedith a pH variation of the dispersing fluid. Generally, acid water

nsat

ent types of kaolin suspensions.

upports the agglomeration of kaolin particles, whereas basicater acts like a deflocculant [19].This tendency is confirmed in the media which were studied

n this work as indicated by the evolution of the electrophore-is mobility as a function of the pH represented in Fig. 10.hese measurements have been carried out using a DELSA 440etameter from Coulter Company.

At low pH surface charge of the particles is small; repul-ion forces are not efficient, then, agglomeration phenomenons strong. At higher pH surface charge is stronger and agglom-ration in the suspension is less important. The evolution of thehase separation index as a function of mass fraction was stud-ed at pH 4 and 11 and compared with results obtained at neutralH (Fig. 11).

The acid suspensions exhibit a strong increase of phase sep-ration index even for the lower values of mass fractions. Thegglomeration phenomenon is strong and seems to be impor-ant even in the lowest particle concentration range which haseen studied. Diluted suspensions domain has disappeared.gglomeration occurs in the same time as sedimentation phe-

omenon takes place even in domain of the lowest concentrationtudied. In fact, the suspensions with a mass fraction of 0.52nd 1.04% are in intermediate state 1. Beyond these concen-rations up to 8% media is in the concentrated suspensions

R. Vie et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 298 (2007) 192–200 199

s as a

dsagc

cpwbpaaaaifsc

3s

wicWmei

viba

Fig. 10. Electrophoresis mobility variation

omain. Solid suspension domain, at pH 4, is reached atlightly lower values than for neutral pH because of strongergglomeration phenomenon. Agglomerates formed are big-er [15,16] and begin to enter in contact at lower particleoncentration.

The low values of the phase separation index for kaolin con-entration of 0.3–12% at pH 11 show that attraction betweenarticles is very weak. Any increase of the slope of PSIhich would indicate concentrated suspension domain cany observed. This shows that at alkaline pH, agglomerationhenomenon is limited. Intermediate zone 2 seems to have dis-ppeared too. In fact, beyond 15% in mass fraction, particlesre sufficiently close to each other to form loose agglomer-tes which in turn form contact network [15,16]. Agglomerationnd network formation occurs nearly at the same time because,

n alkaline media, agglomeration is possible at high massractions. Then, diluted suspensions become solids suspen-ions when mass fraction increases in a very narrow range ofoncentration.

ra

r

Fig. 11. Phase separation index of the different studi

function of the pH of kaolin suspensions.

.4.2. Example of PSI evolution of non-cohesive particleuspensions

PSI evolutions of glass beads suspensions of different sizeere studied as a function of mass fraction as it is represented

n Fig. 11. In the range of sphere radius studied, PSI valueshange with particle size but its evolution remains the same.ithout any agglomeration phenomenon PSI evolution is deeplyodified. It increases linearly from 0.4 to 48% and does not

xhibit any discontinuity. Concentrated suspension domain andntermediate state 1 are not observed.

Beyond 75% samples were comparable to wet sand with aery high viscosity and it was impossible to implement the exper-ments. Unlike the kaolin suspensions, contact network in glassead suspensions is composed of rigid particles and not loosegglomerates. This explains why solid suspension domain is

eached at so high concentration with glass beads suspensionsnd why viscosity is so strongly increased.

Between 50 and 70% in mass fraction PSI behaviour is noteally linear especially for suspensions made from 0–50 �m

ed suspensions as a function of mass fraction.

2 hysic

ga

4

acDc

peottppat

atlItm

pSmptis2ioo

R

[

[

[

[

[

[

[

[

[18] L. Bergstrom, Shear thinning and shear thickening of concentrated ceramic

00 R. Vie et al. / Colloids and Surfaces A: P

lass beads. This is due to the backscattering signal which arrivest saturation at these high concentrations.

. Conclusion

For some critical concentrations properties of suspensionsre abruptly modified due to a change in their interparti-le interactions as it have been shown with clarification rate.estabilisation properties are not modified so sharply because

oncentration evolves while settling processes.Phase separation index which depends on average size of

articles and solid concentration satisfactorily describes thevolution of these destabilisation phenomena as a functionf mass fraction. Starting from variations of the PSI, thehree types of suspensions described in scientific press andwo intermediate zones could be distinguished. It make alsoossible to give discriminant values for all the types of sus-ensions from 0.1 to 40% in mass fraction for model samplesnd up to 70% to industrial samples which contain surfac-ants.

Through results obtained for non-cohesive glass beads, kaolinnd the influence of the pH it has been shown that PSI is sensi-ive to particles granulometry, chemical nature of the solid phase,iquid phase composition and especially particles interactions.t can be proposed, then, to study efficiency of the formula-ion of a suspension or evaluate agglomeration tendency of a

edium.PSI results are supplemented by rheological tests which made

ossible higher mass fractions in model media to be studied.uspensions of kaolin can be characterized by viscous floweasurements in the concentration range from 30 to 55%. K

arameter from Oswald de Weale can be related to mass frac-ion in the medium. Its evolution does not indicate any changen particle organisation after the transition to the solid suspen-ion domain identified from stability study. Thus, Turbiscan MA

000 is suitable to indentification and characterisation of stabil-ty of suspensions. These results make possible a classificationf these complex fluids which is principally based on the naturef particles interactions.

[

ochem. Eng. Aspects 298 (2007) 192–200

eferences

[1] P. Coussot, C. Ancey, Rheophysique des pates et des suspensions, EDPSciences, Les Ulis, 1999.

[2] Th.F. Tradros, Correlation of viscoelastic properties of stable and floccu-lated suspensions with their interparticle interactions, Adv. Colloid Interf.Sci. 68 (1996) 97–200.

[3] B. Fitch, Sedimentation of flocculent suspensions: state of the art, AIChEJ. 25 (6) (1979) 913–930.

[4] D.A. Dahlstrom, B. Fitch, Thickening, in: N.L. Weiss (Ed.), SME Process-ing Handbook, AIME, New York, 1985, pp. 9-2–9-14.

[5] R.G. Holdich, G. Butt, Compression and channelling in gravity sedimentingsystems, Miner. Eng. 9 (1) (1996) 115–131.

[6] P.E. Rosenberg, Am. Miner. 87 (2002) 103–107.[7] J. Larray, Recherche d’une methode de dosage quantitative des kaolinites

et des illites par diffraction X, thesis, Universite Pierre et Marie Curie-ParisVI and Ecole Nationale Superieure des Mines de Paris, 1981.

[8] O. Mengual, G. Meunier, I. Cayre, K. Peuch, P. Snabre, TURBISCAN MA2000: multiple light scattering measurement for concentrated emulsion andsuspension instability analysis, Talanta 50 (2) (1999) 445–456.

[9] A. Papo, L. Piani, R. Ricceri, Sodium tripolyposphate and polyphosphateas dispersing agents for kaolin suspensions: rheological characterization,Colloids Surf. A 201 (2002) 219–230.

10] J. Gregory, The density of particle aggregates, Water Sci. Technol. 36 (4)(1997) 1–13.

11] B. Fitch, Sedimentation process fundamentals, Trans. Am. Inst. Miner. Eng.223 (1962) 129–137.

12] J.R. Karl, S.A. Wells, Numerical model of sedimentation/thickening withinertial effects, J. Environ. Sci. Eng. 125 (9) (1999) 792–805.

13] R. Font, M.L. Laveda, Desgn method of continuous thickener from semi-bath tests of sedimentation, Chem. Eng. Sci. (1996) 5007–5015.

14] P. Sandkuhler, J. Sefcik, M. Morbidelli, Kinetics of gel formation in dilutedispersions with strong attractive particle interactions, Adv. Colloid Interf.Sci. 108–109 (2004) 133–143.

15] P. Snabre, P. Mills, I. Rheology of weakly flocculated suspensions of rigidparticle, J. Phys. III France 6 (1996) 1811–1834.

16] X. Jia, D.J. Wedlock, R.A. Williams, Simulation of simultaneous aggrega-tion and sedimentation, Miner. Eng. 13 (2000) 1349–1360.

17] Q. Huang, W. Li, M. Gu, P. Chen, H. Zhang, Y. Jin, Effect of blend ratioon rheological properties of aqueous SiC suspensions, J. Eur. Ceram. Soc.24 (2004) 2157–2161.

suspensions, Colloids Surf. A 133 (1998) 151–155.19] O. Cuisset, Proprietes electrocinetiques des particules argileuses. Appli-

cation de la methode electrophoretique aux problemes d’environnement etd’identification des sols, Laboratoire Central des Ponts et Chaussees, 1980.